The original version of vmm386 was a fantastic dos multitasker, even to the point of allowing multiple dos programs write to the screen via bios calls.  I wanted to use it for some things back when it was new shareware, and I even called and got to talk to the developer.  The problem for me was that my screen reader intercepted the dos interrupts for writing to the screen, so when I was using vm386, the screen reader would talk both programs simultaneously.  Needless to say, that made for rather confusing operations.  It worked just fine though with programs that bypassed the bios calls, and just wrote to video memory, since those my screen reader didn't intercept.  On the other hand, it made the programs harder to use, so I didn't spend a lot of time using vm386 myself, but I did hear of others using it for all kinds of things, including running bbs software to handle multiple lines.

I know they've tried to make an openssource version of it not so long ago, but that's not from the original sources so far as I could tell, so I don't believe it will/does work quite as well as the original program does.

On the other hand, there's nothing stopping someone from grabbing a copy and installing it and running it on a machine nowadays, since the folks that made it started the opensourcing process, I doubt anyone would complain if someone used it these days.


On 3/28/2023 7:17 PM, Rugxulo wrote:
Hi,

On Tue, Mar 28, 2023 at 8:47 AM Liam Proven <lpro...@gmail.com> wrote:
DR DOS does have some source code available, and includes TaskMaster,
which can do full-screen multitasking of DOS sessions. This *does*
work on bare modern hardware in my testing.
IIRC, DR-DOS 7.03 (circa 1999) had task swapping for 286s and
preemptive multitasking for 386s (TASKMGR.EXE). But you had to use
their DR EMM386.EXE (no HIMEM.SYS needed) with their built-in DPMI
enabled. (It had a lot of bundled / hidden .VXDs or whatever.) It was
limited to 64 MB per task (despite the false claim of XMS v3 support).
And no FAT32 support.

Lineo/DeviceLogics president and CEO Bryan Sparks said all CP/M
derivatives are free to use, modify and distribute last year. DR DOS
is a derivative of CP/M-86 which is a derivative of CP/M. I think it
could be used.
They stopped selling DR-DOS online back in 2018, right? But I'd be
surprised if DR-DOS was still considered a true derivative of CP/M-86.
Almost all of the CP/M support was probably stripped out. I'm overly
skeptical about that. (The so-called "OpenDOS" was only kernel and
shell for "non-commercial use", AFAIK, and wasn't even patched with
the latest Novell fixes.)

It seems to me that if the sources of Multiuser DOS could be obtained,
and if it's covered by Mr Sparks' edict, then it would give a lot of
what people want from a DOS nowadays.
Minix 2.0.4 (circa 2003) could run atop FAT16 (e.g. DOS). It wasn't
perfect but still quite good. It could multitask its own binaries
(a.out variant). I've been wanting to try to build 8086tiny (ecm's
fork) under it. But even Minix choked on machines with lots of RAM. I
don't think it booted atop FAT32 either. I personally wanted to try
again under VirtualBox one of these days.

Or just develop in standard C (or Modula-2) atop Minix [DOSMinix,
booting atop FAT], with its multitasking for faster development, and
later transfer your sources to DOS to compile natively.

You could also run old Slackware 11 (ZipSlack) atop FAT (Linux 2.4
kernel, UMSDOS). IIRC, it had GCC 3.4.6. Maybe even an old DOSEMU
would run there.

Multiuser DOS was the last and final descendant of CP/M. It's a native
32-bit OS, multitasking but DOS compatible, with FAT32 support. It
supports up to 4GB of RAM and apps can get both EMS and XMS services.
Memory is such a mess (and I don't mean 16-bit). So many things have
corner cases or bugs.

In case it wasn't obvious, I did buy DR-DOS (online in 2004), but I
rarely used their multitasking. The main potential uses (to me) would
be 1). finding files in the background (or grepping), 2). compiling
some sources, or 3). file compression. But I rarely needed to care.
(Most people would also prefer listening to music or downloading
files.)

As a workaround, locally in FreeDOS, I always (weakly) tried to
simplify things (build processes), use speedy tools, better
algorithms, etc. Running atop RAM disk and/or cache also helps a ton.
DJGPP can be quite slow (and worse with LFNs enabled). You know, if
everything is quick and efficient (and accurate), you don't need to
multitask as much. (But I hate brittle makefiles that are easy to
break. I'd rather just rebuild slowly from scratch via shell script.)

It has modest hardware support: CD, DVD, sound, mouse, a few other
things. It supports a few network cards, and can talk TCP/IP and SMB.
There are some brilliant apps that use the mouse (e.g. JED), but I
rarely relied on it. Sound is the weakest link in DOS (and probably
not crucial to "real work" for most people). Network can be very
useful but isn't well-supported (lack of packet drivers).

That vaguely reminds me. I think I once suggested someone use FreeBSD
and QEMU as a sort of way to multitask DOS. You don't even need X11
installed. The minimum (last I checked) for FreeBSD was 64 MB of RAM
(486 DX or better), but of course probably much more required with a
guest running. (They've had their own hypervisor, bhyve, since 2014 or
so, using VT-X [EPT], but I don't specifically know if they ever
bothered running DOS with it. I think they did have some shims for
BIOS-based Windows. But stick with the QEMU package for now.)

It's not a true DOS, it can't run DOS device drivers, and has
functionality that's irrelevant today, such as serial terminal
support. But if someone could chase down a final version of the
source, it could have some obsolete stuff stripped out (NetBIOS and
IPX/SPX support, RS/232 terminals, etc.) and could be useful to
someone somewhere.

I tried to contact 3 or 4 vendors of Multiuser DOS mentioned in my
article. Most didn't reply.
Even Minix 3, formerly with lots of funding, still dried up in 2016.
It's sad, but most people don't want a 32-bit only OS that doesn't
have USB support (very complex). Well, except Intel for its Management
Engine.  ;-)   I was always impressed by Minix and how much they
accomplished, even in the 2.x days.

I'm sure there are dozens of improvements we could make to FreeDOS. Or
just workarounds. I just don't have any perfect answers.


_______________________________________________
Freedos-user mailing list
Freedos-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freedos-user


_______________________________________________
Freedos-user mailing list
Freedos-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freedos-user

Reply via email to